[optical-networks] Meeting Time?

[Joe Bannister]

The morning of next Thursday would be best for me.  Either at ISI or UPC.

> Hello everyone!
> 
> Let's get together and have a meeting to discuss the optical correlator
> project.  John and I will be attending from USC.  We are available on:
> (in order of preference) Tuesday morning, Monday after 1pm, and Thursday
> before 4pm.  Let me know if any of these times are good for you (the ISI
> folks) so we can settle on a meeting time.  Also, we can either meet
> here or at ISI, it's up to you.
> 
> We would like this to be a working meeting in which we finalize what we
> want to present in the short (3 page) paper that we would like to submit
> to the Optical Fiber Communications conference (OFC), for which the
> deadline is October 10th.  Since we only have a month left, my hope is
> that after the meeting it is clear what each of us needs to do to get
> this paper done on time (what data do we want to present and so forth).
> After the conference we will submit a paper to Photonics Technology
> Letters in which we can include any extra data that didn't make it into
> conference submission.
> 
> As you know, John and I have been constructing the experimental setup to
> demonstrate an optical correlator that can be dynamically configured to
> "match" any x of 8 bits.  The actual implementation would require an x
> of 24 bits correlator, so the 8 bit one is just to demonstrate that it
> works.  The idea is to perform a hash on a routing table that would
> allow the optical hardware to determine which output port of an IP
> router that an incoming packet should be routed to.  The theory is that
> this can be done by looking at only a subset of the 24 bits in the IP
> header since most internet routers only have 2 to 4 ports to choose
> from.  ISI's part of the research (as I understand it) is to use some
> actual routing tables to determine how successful this might be.  For
> instance, how often would examining just a few of the header bits
> succeed in making a routing decision?  Clearly, for cases in which the
> correlator fails to make a "match," the packet will have to be
> electronically detected and processed in more detail.  Also, the
> correlator must be tunable since which bits you need to look at may
> change as the routing table evolves.
> 
> -Michelle
> 
> 
> _______________________________________________
> Optical-networks mailing list
> Optical-networks@mailman.isi.edu
> http://mailman.isi.edu/mailman/listinfo/optical-networks